The extraordinary problem with a perfectly ordinary food label

Walk into almost any supermarket and the nutritional label will tell us how much fibre a food contains, expressed in grams. It seems a satisfyingly exact piece of information. A cereal contains 8 grams, a portion of lentils contains 5 grams and an orange contains rather less. We add the figures together and congratulate ourselves on a day well spent looking after our intestines.

Unfortunately, a gram of one fibre need not behave like a gram of another. A whole apple and a puréed apple can contain much the same fibre yet differ in the way we eat and digest them. A coarse wheat bran and the same bran milled into a finer powder can produce different stool weights. Even a bowl of oats can produce a different blood glucose response according to how the grain has been processed. Fibre is not one substance, and the label is an excellent starting point rather than the last word.

To understand why, we need to follow the relationship between carbohydrates, fibre, the physical architecture of food and the glycaemic index. It sounds a formidable excursion. Happily, the chemistry begins with something surprisingly familiar: sugar.

Fibre is mostly carbohydrate, but not all carbohydrates behave alike

Carbohydrates include sugars, starches and most dietary fibres. Their molecules are built from smaller chemical units, frequently sugars, joined by chemical bonds. Starch and cellulose provide a particularly elegant demonstration: both are made from glucose, yet the arrangement of the bonds is different. Humans possess enzymes that readily break many of the alpha linkages in starch. We do not possess the equivalent enzymes for the beta-1,4 linkages of cellulose.

As a result, much ordinary starch is converted to absorbable glucose in the small intestine, whereas cellulose remains largely inaccessible to our own enzymes. Other fibre compounds, including pectin and inulin, also evade digestion by our enzymes but can be fermented by microbes living in the large intestine. These microbes produce gases and short-chain fatty acids, including acetate, propionate and butyrate, that have physiological roles of their own.

There is an interesting borderline case called resistant starch. It is starch, chemically speaking, yet part of it escapes digestion in the small intestine because of its molecular arrangement or the physical structure enclosing it. In UK nutrition policy, resistant starch forms part of dietary fibre. The apparently tidy division between 'starch' and 'fibre' therefore turns out to be not quite so tidy after all.

For UK nutritional purposes, digestible carbohydrate is conventionally assigned 4 kilocalories per gram and dietary fibre 2 kilocalories per gram. The second figure is an average accounting convention reflecting energy that may be obtained, particularly through fermentation. It does not mean that every fibre reliably yields precisely two calories, still less that fibre is simply inert material passing untouched through the body.

Four properties, not four competing varieties

Fibre is frequently introduced as either soluble or insoluble. That remains useful vocabulary, but it is only one dimension of the story. We can also ask how viscous a fibre becomes in water, how readily microbes ferment it and what physical structure it has within a food.

PropertyThe question being askedWhy digestion may be affected
SolubilityDoes the material dissolve or disperse in water under the conditions tested?It helps describe behaviour in the digestive tract, but does not determine that behaviour by itself.
ViscosityHow strongly does it thicken the surrounding liquid?Some gel-forming fibres slow access to and absorption of nutrients.
FermentabilityTo what extent and at what rate can intestinal microbes use it?Fermentation produces short-chain fatty acids and sometimes troublesome gas.
Physical structureHow large are the particles, how intact are the plant cells and how long are the fibre molecules?Structure affects chewing, water retention, enzyme access, movement through the bowel and sometimes microbial access.

These are not four percentages of a single total. We cannot sensibly describe a food as 22 per cent soluble, 28 per cent viscous, 32 per cent fermentable and 18 per cent structural and then add the result to 100. Solubility can be measured as a fraction; viscosity might be expressed in physical units; fermentability depends on the microbes and time allowed; structure requires several measurements. The properties overlap. Pectin may be soluble and fermentable. Psyllium can form a viscous gel while being comparatively resistant to rapid fermentation. Wheat bran generally has a very different relationship with water and stool bulk.

These distinctions are reflected in guidance from the British Nutrition Foundation, which cautions against relying on the soluble versus insoluble division alone.

What do we mean by the 'length' of fibre?

There are at least three lengths worth separating. First is the molecular chain length, meaning the number of linked units in a carbohydrate molecule. Scientists often express this indirectly as molecular weight. Second is particle size, meaning the dimensions of the pieces of grain, fruit or vegetable remaining after processing. Third is the extent to which plant cells and cell walls remain intact, creating a physical barrier around starches, fats and other nutrients.

These are different things. A kitchen blender can shred celery into particles too small for us to recognise, without necessarily snapping the individual cellulose molecules into shorter chemical chains. A mill can turn a grain into fine flour without removing its fibre. By contrast, certain more intensive manufacturing operations, enzymes or chemical conditions can change the molecular size of fibres such as oat beta glucan, potentially altering the viscosity that helps determine their effects.

Think of a bundle of string. We might chop the bundle into small sections without cutting every individual strand into shorter molecules. Or we might actually shorten the strands themselves. To the eye both processes can look like 'breaking down fibre', but the consequences are not necessarily identical.

Whole, crushed and fully blended: what does the evidence suggest?

The following table is a comparison of general tendencies, not a promise about every food. There are good experiments comparing intact with extensively processed foods. Direct trials of the intermediate category, lightly crushed or briefly blended, are less common. The middle column is therefore partly an informed expectation rather than a separately established clinical result.

FeatureWhole or largely intactChopped, crushed or lightly blendedFinely blended or milled
Total fibre, if nothing is discardedRetainedGenerally retainedGenerally retained
Plant cell integrityOften greatestPartially disruptedFrequently substantially disrupted
Need to chewUsually greatestUsually reducedOften minimal
Access of digestive enzymes to starchMay be restricted by intact cellsCan increaseOften greater, especially in cereals
Blood glucose responseOften moderated in intact grainsFood-specific, sparsely tested as a distinct categoryCan be higher in milled cereals; not reliably higher in blended fruit or pulses
Fullness and eating rateOften favours fullness and slower eatingIntermediate effects are plausibleSome fruit purées and drinks are less filling, but context matters
Microbial fermentationDepends on chemistry and accessAccessibility may changeMay change rate or site of fermentation; direction is not universal
Stool bulkingCoarse bran can be particularly effectiveNot well quantified across foodsFine bran can be less effective than coarse bran; other fibres behave differently

One further distinction is essential: blending is not juicing. A blender usually retains the fruit flesh, peel and fibre placed in it. A juicer that separates and discards pulp can remove a substantial proportion of that material. Sieving a smoothie afterwards can also change the composition. Comparing whole fruit with filtered juice does not isolate the effect of blending alone.

Apples, berries and the danger of a universal smoothie rule

Some of the most memorable research began with the humble apple. In 1977, Haber and colleagues fed ten participants whole apples, apple purée and apple juice containing comparable available carbohydrate. The drinks and purées were consumed much faster than the intact fruit, and the researchers reported differences in fullness and later blood glucose behaviour. A larger study published in 2009, involving 58 adults, also found that whole apple was more filling than apple sauce or juice when eaten before a meal, even with several characteristics of the test portions controlled.

The results support an important idea: physical form can matter, especially when it changes chewing, eating speed and satisfaction. But we would be making a mistake if we turned 'whole fruit can be more filling' into 'blending fruit always makes its sugar more damaging'.

In a 2022 crossover experiment with 20 healthy young adults, blended Gala apple and blackberries produced a lower measured post-meal glucose response than the same fruits eaten whole. The researchers reported a lower peak glucose rise and a lower incremental area under the glucose curve. It is a small study using particular fruits in a particular population, not proof that all smoothies improve blood glucose control. It is, however, decisive evidence against treating the opposite proposition as an unquestionable law.

Rather than issuing blanket instructions about blending, the sensible question is what fruit, how much, what else is added, how quickly it is consumed and which outcome we care about.

Grains are where food structure becomes particularly convincing

For cereals, there is stronger evidence of a processing effect on the absorption of carbohydrate. Oat grains contain starch embedded in a plant structure alongside beta glucan, a fibre capable of increasing viscosity. Cutting, flaking, cooking and instantising can change the access of digestive enzymes to the starch, even while the food remains, in a legitimate sense, wholegrain.

A 2015 systematic review combined published and database measurements of glycaemic index for oat foods. It reported mean GI values of about 53 for large-flake oats, 55 for steel-cut oats, 71 for quick-cooking oats and 75 for instant oatmeal. These are averages across particular tested products, not fixed values for every packet on a supermarket shelf. The authors identified smaller particle size and greater starch gelatinisation as likely contributors. A subsequent meta-analysis of randomised trials likewise found that intact oat kernels and thicker flakes reduced glucose and insulin responses relative to refined grain controls, while thinner or instant forms did not show the same advantage.

The lesson is not that instant oats have become nutritionally worthless. Their fibre has not vanished. Rather, the surrounding starch has become more accessible. The number of fibre grams alone cannot reliably predict the glucose curve.

There is another subtlety. Oat beta glucan's molecular weight and viscosity may matter, not merely the amount present. Some controlled experiments support this mechanism; others find that altering measured viscosity by a particular preparation technique does not necessarily change the overall two-hour glucose response. That disagreement is scientifically useful: a plausible mechanism still has to survive a test in real meals and real people.

Coarse bran, lentils and almonds complicate the picture further

Stool bulk gives us a different outcome to investigate. In a 1978 experiment reported in Gut, 21 healthy volunteers participated in 28 comparisons of coarse and finely milled wheat bran. With 20 grams of bran per day, mean stool weight was about 219 grams per day with coarse bran and 199 grams with fine bran. The coarse material also had greater measured water-holding capacity. Here physical particle size made an observable difference despite broadly comparable fibre material. This is an older and relatively small study, so it should not be used to predict individual bowel results to the nearest gram.

Pulses provide a useful caution against assuming that flour must always be worse. In a 2014 set of randomised experiments, researchers compared whole, puréed and powdered navy beans, lentils and chickpeas. Processing did not reliably predict the blood glucose response. Several processed pulse treatments retained advantages compared with a whole-wheat flour control. The surrounding food chemistry and composition continued to matter.

Almonds offer a similar counterexample for bowel function. A 2022 trial randomly assigned 87 adults to consume whole almonds, ground almonds or a control snack for four weeks. Grinding changed particle size and predicted lipid release, but almond form did not significantly alter measured gut transit time, stool consistency or gut symptoms. The study did find some differences when almonds, as a group, were compared with control, including greater faecal butyrate concentration. The result is a useful antidote to the assumption that every mechanical change must produce a measurable clinical benefit or harm.

Seeds also deserve distinction. Breaking a linseed's hard coat can expose its contents, but that fact alone does not establish that finely ground linseed is a better treatment for constipation than intact seeds in every person. Improved nutrient availability, bowel movement frequency and blood glucose regulation are separate questions requiring separate evidence.

Now the other half of the story: glycaemic index

The glycaemic index, or GI, measures how much a food raises blood glucose over a standard testing period compared with a reference, usually glucose. Standard testing usually compares portions containing 50 grams of available carbohydrate, with a 25-gram allowance for certain foods where a 50-gram portion would be impractically large. It is an experimentally measured response, not a number calculated from the four fibre characteristics.

Fibre can influence that response in several ways. Viscous fibre may slow the movement and absorption of nutrients. Intact cell walls can protect starch from digestive enzymes. Fermentable material can have effects further along the digestive tract, although fermentability is not a direct shortcut to estimating a food's GI. But GI also depends on the type and accessibility of starch, cooking, cooling, ripeness, acidity, protein, fat, portion preparation and the testing method. Different people can respond differently to the same meal.

GI tells us relatively little about the amount of carbohydrate in the serving actually eaten. That is where glycaemic load, or GL, becomes useful:

Glycaemic load = glycaemic index × grams of available carbohydrate in the portion ÷ 100.

A serving with GI 70 and 10 grams of available carbohydrate has GL 7. Another with GI 50 and 40 grams has GL 20. The second food has the lower GI but the higher glycaemic load. These are illustrative calculations, not measured values for a particular food.

A 2021 international systematic review assembled GI values for more than 4,000 food entries and found substantial variation among cereals and breads, including wholegrain products, whereas legumes and fruit were more consistently lower GI. That variation is exactly what we should expect when food composition, manufacturing and structure all contribute.

Cooking and cooling: carbohydrate changes without a blender

Mechanical processing is only one way to alter starch. Cooking generally hydrates and gelatinises starch, often making it more accessible to digestive enzymes. During cooling, some starch molecules can reorganise into forms that resist digestion, a process known as retrogradation. This is part of the explanation for resistant starch in some cooled cooked foods.

In a randomised crossover experiment involving 30 women with elevated fasting glucose and insulin, chilled potatoes produced lower glucose and insulin concentrations at some early time points than freshly boiled potatoes. The investigators did not find a significant difference in the overall glucose area under the curve. That is an important qualification: a change at 15 or 30 minutes is not necessarily a large improvement across the entire period, nor does cooling magically transform potatoes into an unrestricted food.

Here, again, carbohydrate chemistry, preparation and physical structure are interacting. No fibre label reports all of them.

And what about constipation, IBS and everyday digestion?

We have been discussing blood glucose, but bowel function is a separate outcome. A fibre that blunts a glucose spike need not be the best at increasing stool weight, and a fibre that promotes fermentation may also cause bloating or discomfort in someone sensitive to gas production.

The British Dietetic Association's evidence-based guidance for chronic constipation distinguishes between individual fibres rather than treating them as interchangeable. Its review supports psyllium for improving stool frequency and consistency, while results for several other supplements are weaker or inconsistent. The guidelines also emphasise that some fibres, notably inulin-type fructans, can increase flatulence. Evidence quality varies between particular interventions and outcomes.

For people with irritable bowel syndrome, this matters even more. Someone might tolerate gel-forming psyllium but find a rapid increase in wheat bran or highly fermentable ingredients uncomfortable. 'More fibre' is not a complete prescription. Type, amount, fluid intake, pace of change and the symptom being addressed all matter.

For the general UK population, the Scientific Advisory Committee on Nutrition recommends around 30 grams of dietary fibre per day for adults, using the relevant analytical definition. This is a population target supported by broader evidence linking higher fibre intakes, particularly from wholegrains and other fibre-rich foods, with better long-term health outcomes. It is not a claim that exactly 30 grams will optimise every individual's digestion or that all sources are interchangeable.

A more useful way to compare the food on our plates

Instead of asking only, 'How many grams of fibre does this contain?', we could reasonably ask five questions. How much fibre is there? What kind of physical and chemical behaviour does it have? How intact is the food's structure? What happens to its available carbohydrate after preparation? And what outcome are we actually interested in: glucose regulation, fullness, stool consistency, comfort or the long-term health of the gut?

A pear is not a celery stick, and neither is a bowl of lentils or finely milled wheat flour. They differ in fibre chemistry, water content, cell structure, starch composition and the way we consume them. The differences are real. What we must resist is the temptation to reduce them to a new simplistic rule, such as 'long fibre good, short fibre bad' or 'never blend fruit'.

The central conclusion is both useful and modest: the grams of fibre on the packet tell us how much material has been measured, but not everything about what that material will do. Its chemistry, physical structure and preparation can alter digestion. The size of the effect depends on the food, the person and the outcome measured. Counting fibre remains worthwhile. It simply should not be mistaken for understanding fibre.